Announcement

Collapse
No announcement yet.

Disk Test (testing) using NAS storage

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Disk Test (testing) using NAS storage

    Hi, I am a (happy) Passmark Performance test user. In my copy of Performance Test, I can select the drive I want to use for the Disk Test subsection of Performance Test.
    That was very cool back when I was trying to convince myself that SSDs were ready for prime time, and worth the investment. (thanks Passmark)
    Having gotten my primary servers tuned up, I have turned my attention to NAS. Immediately I noticed the NAS products (multi-bay boxes attached via ethernet) are certainly not created equal.
    I'm not one to go shooting in the dark so I'm thinking about what I want to measure as I compare the various NAS products and their options/configurations, etc.

    My first thought was to run the tried and true Passmark tests, but alas, the Disk Test only allows me to select directly attached drives. Shucks, that would have been so easy !

    I can argue both ways regarding the correctness of the decision to limit the Disk Test to direct-attached drives. I might also argue that a meaningful NAS test would be accomplished via multiple network clients versus from a single server. (i.e. the Performance/Disk test is simply the wrong approach)
    I recognize Passmark and this forum as thought leaders in the testing field, thus,
    I would be curious to know if the forum feels like including remote drives as a "choice" in the Disk Test makes sense.

    I understand a request of this sort is tangential to the purpose for which Performance Test serves so well- global baseline comparison. Given that, what other approach might one take to accomplish baseline comparisons for remote disks, specifically NAS boxes?
    I'm going to suggest, certain NAS devices are fast enough (assuming a great GPU) to accomodate all but the most demanding post production rendering, CAD, games, etc.
    But which ones? (smile)

    Thanks for your indugence. Any and all discussion would be greatly appreciated.








  • #2
    There is not a limitation of testing only local drives.
    What you are seeing is likely a dumb Windows quirk that appears to cause this, but there is a work around!

    You can map a drive in Windows. i.e. create a drive letter for a remote drive. This can any storage really. A NAS box, remote file share, cloud drive, RAM drive, Truecrypt virtual drive, etc...

    Once you have mapped a drive letter, then it will appear in PerformanceTest.

    BUT and this is a big BUT. Windows is dumb and if you map a drive as the Elevated Admin user, then it doesn't appear for other users. And if you map a drive as a normal user, then it doesn't appear for the Admin users. And PerformanceTest only runs as the Elevated Admin user.

    So you either need to map the drive as Elevated Admin. Or do the following registry hack.

    Edit/create the "EnableLinkedConnections" registry value at "HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows \Cur rentVersion\Policies\System" and set it to a value of 1.

    There is more information available here.

    Registry hack
    Click image for larger version  Name:	Linking-mapped-drives.png Views:	0 Size:	53.7 KB ID:	54513
    .
    Mapping a drive letter in Windows
    You can right click on a folder in Windows explorer to map a NAS folder to a drive letter.
    More details on mapping drives is here.
    Click image for larger version  Name:	map-drive.png Views:	0 Size:	17.1 KB ID:	54514

    Then in PerformanceTest you can see all the drive letters.
    C: is local
    D: is local
    S: Is virtual drive
    Z: Is remote file share / NAS.
    Click image for larger version  Name:	Drive-Performance.png Views:	0 Size:	24.0 KB ID:	54515


    But in our experience the performance of local drives is hugely better than remote drives. The unavoidable network latency kills the performance.
    The numbers above (0.29MB/sec) was the performance of our network file server across a decent internet connection via a VPN. If tested locally the same drive is 1500x faster.







    Comment


    • #3
      David,
      Man, Thank you. That was a great response. Great news about the remote drives. And hey, I surely did not mean to drag you into the Windows-weeds. I have tried all the elevated combinations - think Im going to go for the reg hack.
      About my tests, and thanks for sharing your data, I already know the local disks will beat the network disks. I'm just happy to have something I trust to compare NAS products to other NAS products to begin with. Performance Test by default will be catching my baselines, so if you are interested, you know where to find them

      Comment

      Working...
      X